0% found this document useful (0 votes)
7 views10 pages

Poisson Distribution Analysis and Applications

This paper presents an in-depth analysis of the Poisson Distribution, covering its historical background, properties, and real-life applications in various fields such as traffic flow and fault prediction. It discusses the relationship between the Poisson and binomial distributions, as well as extensions like the Zero-Inflated Poisson and Negative Binomial distributions to address limitations like overdispersion. Overall, the Poisson distribution is emphasized as a crucial tool for statistical modeling of count data in both scientific and industrial contexts.

Uploaded by

Ayush Kumar
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
7 views10 pages

Poisson Distribution Analysis and Applications

This paper presents an in-depth analysis of the Poisson Distribution, covering its historical background, properties, and real-life applications in various fields such as traffic flow and fault prediction. It discusses the relationship between the Poisson and binomial distributions, as well as extensions like the Zero-Inflated Poisson and Negative Binomial distributions to address limitations like overdispersion. Overall, the Poisson distribution is emphasized as a crucial tool for statistical modeling of count data in both scientific and industrial contexts.

Uploaded by

Ayush Kumar
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

‭An Improved Assessment of Poisson’s Distribution and Applications‬

‭Himpriya Kumari and Ashok Pal‬


‭Department of Mathematics, Chandigarh University, Gharuan, Mohali - 140413‬
‭Email address:‬‭24msm10007@[Link]‬‭,‬‭[Link]@[Link]‬

‭ABSTRACT‬

‭ his‬ ‭paper‬ ‭provides‬ ‭a‬ ‭comprehensive‬ ‭analysis‬ ‭of‬ ‭the‬ ‭Poisson‬ ‭Distribution,‬ ‭including‬ ‭its‬ ‭historical‬
T
‭background‬ ‭and‬ ‭past‬ ‭developments.‬ ‭We‬ ‭examine‬‭its‬‭properties‬‭and‬‭explore‬‭its‬‭real-life‬‭applications,‬
‭discussing‬ ‭key‬ ‭concepts‬ ‭such‬ ‭as‬ ‭mean,‬ ‭variance,‬ ‭and‬ ‭moments.‬ ‭Additionally,‬ ‭we‬ ‭delve‬ ‭into‬ ‭the‬
‭Poisson‬‭process,‬‭its‬‭properties,‬‭and‬‭its‬‭applications.‬‭The‬‭Poisson‬‭Distribution‬‭is‬‭particularly‬‭useful‬‭in‬
‭analyzing‬‭scenarios‬‭like‬‭traffic‬‭flow,‬‭predicting‬‭faults‬‭in‬‭electric‬‭cables,‬‭and‬‭estimating‬‭the‬‭occurrence‬
‭of‬‭random‬‭accidents.‬‭It‬‭is‬‭especially‬‭relevant‬‭in‬‭cases‬‭where‬‭events‬‭are‬‭rare‬‭within‬‭a‬‭large‬‭sample‬‭or‬
‭observation‬‭space.‬‭Furthermore,‬‭there‬‭is‬‭a‬‭close‬‭relationship‬‭between‬‭the‬‭Poisson‬‭distribution‬‭and‬‭the‬
‭binomial‬ ‭distribution,‬ ‭where‬ ‭the‬ ‭former‬ ‭can‬ ‭be‬ ‭derived‬ ‭as‬ ‭an‬ ‭approximation‬ ‭of‬ ‭the‬ ‭latter‬ ‭when‬ ‭the‬
‭number‬‭of‬‭trials‬‭is‬‭large‬‭and‬‭the‬‭probability‬‭of‬‭success‬‭is‬‭small.‬‭The‬‭paper‬‭also‬‭addresses‬‭extensions‬‭of‬
‭the‬‭Poisson‬‭model,‬‭such‬‭as‬‭the‬‭Zero-Inflated‬‭Poisson‬‭and‬‭Negative‬‭Binomial‬‭distributions,‬‭which‬‭help‬
‭manage‬ ‭limitations‬ ‭like‬ ‭overdispersion‬ ‭and‬ ‭excess‬ ‭zeros,‬ ‭making‬ ‭the‬ ‭model‬ ‭adaptable‬ ‭to‬ ‭more‬
‭complex‬‭data‬‭scenarios.‬‭Overall,‬‭the‬‭Poisson‬‭distribution‬‭remains‬‭a‬‭fundamental‬‭tool‬‭for‬‭the‬‭statistical‬
‭modelling‬ ‭of‬ ‭count‬ ‭data,‬ ‭supporting‬ ‭analyses‬‭in‬‭both‬‭scientific‬‭and‬‭industrial‬‭contexts.‬‭We‬‭will‬‭also‬
‭explore‬ ‭the‬ ‭limitations‬‭and‬‭assumptions‬‭inherent‬‭to‬‭this‬‭distribution‬‭throughout‬‭the‬‭paper,‬‭especially‬
‭focusing‬‭on‬‭its‬‭appropriateness‬‭in‬‭contexts‬‭where‬‭events‬‭are‬‭rare‬‭across‬‭large‬‭observation‬‭spaces,‬‭and‬
‭we will discuss the relationship between the Poisson and binomial distributions in depth.‬

‭ EYWORDS‬ ‭–‬ ‭Probability‬ ‭distribution,‬ ‭binomial‬ ‭distribution,‬ ‭discrete‬ ‭distribution,‬ ‭Poisson‬


K
‭distribution‬

‭1 INTRODUCTION‬

‭ he‬ ‭Poisson‬ ‭distribution‬ ‭is‬ ‭an‬ ‭important‬ ‭concept‬ ‭in‬ ‭probability‬ ‭and‬ ‭statistics,‬ ‭commonly‬ ‭used‬ ‭to‬
T
‭represent‬ ‭the‬ ‭frequency‬ ‭of‬ ‭events‬ ‭occurring‬ ‭over‬ ‭a‬ ‭specific‬ ‭period‬ ‭or‬ ‭area.‬ ‭Developed‬ ‭by‬ ‭Siméon‬
‭Denis‬ ‭Poisson‬ ‭in‬ ‭the‬ ‭early‬ ‭19th‬ ‭century,‬ ‭it‬ ‭was‬ ‭initially‬ ‭designed‬ ‭to‬ ‭analyze‬ ‭rare‬ ‭events,‬ ‭such‬ ‭as‬
‭unlikely‬‭accidents‬‭or‬‭natural‬‭occurrences‬‭within‬‭defined‬‭time‬‭frames.‬‭Over‬‭time,‬‭it‬‭has‬‭evolved‬‭into‬‭a‬
‭vital‬‭tool‬‭for‬‭understanding‬‭and‬‭predicting‬‭event‬‭frequencies‬‭across‬‭various‬‭fields,‬‭including‬‭physics,‬
‭biology, telecommunications, finance, and public health. [5]‬

‭ ne‬‭of‬‭the‬‭key‬‭features‬‭of‬‭the‬‭Poisson‬‭distribution‬‭is‬‭its‬‭applicability‬‭to‬‭situations‬‭where‬‭events‬‭occur‬
O
‭randomly,‬‭independently,‬‭and‬‭at‬‭a‬‭consistent‬‭average‬‭rate.‬‭This‬‭makes‬‭it‬‭particularly‬‭relevant‬‭for‬‭data‬
‭related‬ ‭to‬ ‭infrequent‬ ‭or‬ ‭unpredictable‬ ‭events,‬ ‭such‬ ‭as‬ ‭the‬ ‭number‬ ‭of‬ ‭meteor‬ ‭sightings‬ ‭in‬ ‭a‬ ‭night,‬
‭incoming‬ ‭calls‬ ‭at‬ ‭a‬ ‭call‬ ‭centre,‬ ‭or‬ ‭system‬ ‭failures‬ ‭during‬ ‭a‬ ‭specified‬ ‭time.‬ ‭It‬ ‭is‬ ‭mathematically‬
‭expressed‬‭through‬‭a‬‭single‬‭parameter,‬‭λ‬‭(lambda),‬‭which‬‭represents‬‭both‬‭the‬‭mean‬‭and‬‭variance‬‭of‬‭the‬
‭distribution. [4]‬

‭ espite‬‭its‬‭straightforward‬‭nature,‬‭the‬‭Poisson‬‭distribution‬‭is‬‭based‬‭on‬‭several‬‭assumptions,‬‭including‬
D
‭the‬‭independence‬‭of‬‭events‬‭and‬‭a‬‭constant‬‭rate‬‭of‬‭occurrence.‬‭However,‬‭real-world‬‭data‬‭often‬‭deviates‬
‭from‬ ‭these‬ ‭assumptions,‬ ‭leading‬‭to‬‭the‬‭development‬‭of‬‭extensions‬‭such‬‭as‬‭the‬‭Zero-Inflated‬‭Poisson‬
‭(ZIP)‬ ‭and‬ ‭the‬ ‭Negative‬ ‭Binomial‬ ‭distribution.‬ ‭These‬ ‭adaptations‬ ‭address‬ ‭challenges‬ ‭like‬
‭overdispersion‬ ‭or‬ ‭the‬ ‭presence‬ ‭of‬ ‭excess‬‭zeros,‬‭ensuring‬‭that‬‭the‬‭Poisson‬‭model‬‭remains‬‭relevant‬‭in‬
c‭ ontemporary‬‭statistical‬‭analysis,‬‭especially‬‭in‬‭scenarios‬‭where‬‭there‬‭is‬‭significant‬‭variability‬‭in‬‭event‬
‭frequency. [6]‬

I‭ n‬ ‭conclusion,‬ ‭the‬ ‭Poisson‬ ‭distribution‬ ‭is‬ ‭an‬ ‭essential‬ ‭tool‬ ‭for‬ ‭statistical‬ ‭modelling,‬ ‭bridging‬
‭theoretical‬ ‭principles‬ ‭with‬ ‭practical‬ ‭applications.‬ ‭Its‬ ‭wide‬ ‭range‬ ‭of‬ ‭uses,‬ ‭along‬ ‭with‬ ‭advanced‬
‭extensions,‬ ‭highlights‬ ‭its‬ ‭importance‬ ‭in‬ ‭tackling‬ ‭modern‬ ‭data‬ ‭analysis‬ ‭challenges‬ ‭in‬ ‭areas‬ ‭where‬
‭predicting events and modelling counts are crucial.‬

‭2 LITERATURE SURVEY‬

‭ he‬‭Poisson‬‭distribution,‬‭introduced‬‭by‬‭the‬‭French‬‭mathematician‬‭Siméon‬‭Denis‬‭Poisson‬‭in‬‭1837,‬‭is‬‭a‬
T
‭statistical‬ ‭concept‬ ‭that‬ ‭describes‬ ‭the‬ ‭probability‬ ‭of‬ ‭a‬ ‭certain‬ ‭number‬ ‭of‬ ‭events‬ ‭occurring‬ ‭within‬ ‭a‬
‭specified‬‭timeframe‬‭or‬‭area.‬‭The‬‭origins‬‭of‬‭the‬‭Poisson‬‭distribution‬‭are‬‭linked‬‭to‬‭earlier‬‭developments‬
‭in‬‭mathematics‬‭and‬‭the‬‭growing‬‭interest‬‭in‬‭probability‬‭theory‬‭during‬‭the‬‭18th‬‭and‬‭19th‬‭centuries.‬‭This‬
‭distribution‬ ‭emerged‬ ‭from‬‭Poisson’s‬‭investigations‬‭into‬‭probability‬‭theory‬‭as‬‭it‬‭pertained‬‭to‬‭criminal‬
‭and‬ ‭civil‬ ‭legal‬ ‭cases.‬ ‭His‬ ‭findings,‬ ‭published‬ ‭in‬ ‭"Recherches‬ ‭sur‬ ‭la‬ ‭Probabilité‬ ‭des‬ ‭Jugements‬ ‭en‬
‭Matière‬ ‭Criminelle‬ ‭et‬ ‭en‬ ‭Matière‬ ‭Civile,"‬ ‭represent‬ ‭one‬ ‭of‬ ‭the‬ ‭early‬ ‭attempts‬ ‭to‬ ‭probabilistically‬
‭model‬ ‭the‬ ‭frequency‬ ‭of‬ ‭rare‬ ‭events.‬ ‭The‬‭Poisson‬‭distribution‬‭gained‬‭practical‬‭interest‬‭and‬‭empirical‬
‭support‬ ‭through‬ ‭the‬ ‭work‬ ‭of‬ ‭Ladislaus‬ ‭Bortkiewicz,‬ ‭a‬ ‭Russian‬ ‭statistician‬ ‭of‬ ‭Polish‬ ‭descent.‬ ‭In‬‭his‬
‭1898‬ ‭publication,‬ ‭"Das‬ ‭Gesetz‬ ‭der‬ ‭Kleinen‬ ‭Zahlen"‬ ‭("The‬ ‭Law‬ ‭of‬ ‭Small‬ ‭Numbers"),‬ ‭Bortkiewicz‬
‭analyzed‬‭data‬‭regarding‬‭the‬‭number‬‭of‬‭soldiers‬‭who‬‭died‬‭from‬‭horse‬‭kicks‬‭in‬‭the‬‭Prussian‬‭army‬‭over‬‭a‬
‭specified‬ ‭period.‬ ‭He‬ ‭discovered‬ ‭that‬ ‭the‬ ‭Poisson‬ ‭distribution‬ ‭effectively‬ ‭modelled‬ ‭these‬ ‭rare,‬
‭seemingly‬ ‭random‬ ‭events.‬ ‭Despite‬ ‭the‬ ‭infrequency‬ ‭of‬ ‭horse‬ ‭kick‬ ‭fatalities,‬ ‭Bortkiewicz‬ ‭found‬ ‭that‬
‭they‬‭occurred‬‭at‬‭a‬‭relatively‬‭consistent‬‭rate‬‭over‬‭the‬‭years,‬‭aligning‬‭well‬‭with‬‭the‬‭assumptions‬‭of‬‭the‬
‭Poisson‬‭model.‬‭His‬‭research‬‭helped‬‭validate‬‭the‬‭distribution‬‭as‬‭a‬‭reliable‬‭tool‬‭for‬‭analyzing‬‭rare‬‭events‬
‭and introduced it to a broader statistical audience. [3]‬

‭3 POISSON DISTRIBUTION‬

‭3.1‬‭Definition‬‭-‬‭A‬‭variable‬‭x‬‭is‬‭said‬‭to‬‭have‬‭the‬‭Poisson‬‭distribution‬‭if‬‭it‬‭takes‬‭the‬‭values‬‭0,‬‭1,‬‭2,…‬‭to‬
‭2‬
−‭𝑚‬ ‭𝑚‬ −‭𝑚‬ ‭𝑚‬ −‭𝑚‬
‭with probabilities‬‭𝑒‬ , ‭‬ ‭1‬! ‭𝑒‬ , ‭‬ ‭2‬! ‭𝑒‬ ,...., ‭‬‭𝑡𝑜‬‭‬∞ ‭respectively. (See figure 1.1) [1]‬

‭Note‬‭: Poisson Distribution has only one parameter m.‬

‭.‬

‭Figure 1.1.‬
‭ robability‬ ‭Mass‬ ‭Function‬ ‭(PMF)‬ ‭-‬ ‭A‬ ‭discrete‬ ‭irregular‬ ‭variable‬ ‭is‬ ‭said‬ ‭to‬ ‭have‬ ‭a‬ ‭Poisson‬
P
‭Distribution‬ ‭with‬ ‭boundary‬ ‭m>0,‬ ‭if‬ ‭for‬ ‭x=0,‬ ‭1,‬ ‭2,‬ ‭….‬ ‭at‬ ‭that‬ ‭point‬ ‭the‬ ‭likelihood‬ ‭mass‬ ‭capacity‬ ‭is‬
‭given by:‬
‭𝑥‬ −‭𝑚‬
‭𝑚‬ ‭𝑒‬
‭P(x) =‬ ‭𝑥!‬

‭ here‬ ‭e‬ ‭is‬ ‭Euler’s‬ ‭number‬ ‭(e=2.178…)‬ ‭x‬ ‭is‬ ‭the‬ ‭number‬ ‭of‬ ‭occurrences‬‭and‬‭x!‬‭is‬‭the‬‭factorial‬‭of‬‭x.‬
w
‭[2][6]‬

‭A graphical representation of PMF is shown in Figure number 1.2.. [5]‬

‭Figure 1.2.‬

‭3.2 Mode of Poisson Distribution‬

‭The mode of Poisson Distribution is the value of the variate x which gives the greatest probability.‬

‭ herefore,‬ ‭the‬ ‭probability‬ ‭of‬ ‭(x-1)‬ ‭successes‬ ≤ ‭probability‬ ‭of‬ ‭x‬ ‭successes‬ ≥ ‭probability‬ ‭of‬ ‭(x+1)‬
T
‭successes‬
−‭𝑚‬ (‭𝑥‬−‭1‬) −‭𝑚‬ ‭𝑥‬ −‭𝑚‬ (‭𝑥‬+‭1)‬
‭𝑒‬ ‭𝑚‬ ‭𝑒‬ 𝑚 ‭ ‬ ‭𝑒‬ ‭𝑚‬
‭i.e.,‬ (‭𝑥−
‬ ‭1‬)!
≤ ‭𝑥‬!
≥ (‭𝑥+
‬ ‭1‬)!

−‭1‬
‭𝑚‬ ‭1‬ ‭𝑚‬
‭or,‬ (‭𝑥‬−‭1‬)! ≤ ‭𝑥‬{(‭𝑥−
‬ ‭1‬)!}!
≥ (−‭1‬)‭𝑥{‬ (‭𝑥−
‬ ‭1)‬ !}

−‭1‬
‭𝑚‬ ‭‬
1 ‭𝑚‬
‭or,‬ ‭1‬
≤ ‭𝑥‬
≥ (‭𝑥+
‬ ‭1‬)‭𝑥‬

‭These give‬‭𝑚‬ − ‭1‬ ≤ ‭𝑥‬ ≤ ‭𝑚‬

I‭ f‬‭m‬‭is‬‭not‬‭an‬‭integer‬‭then‬‭the‬‭mode‬‭is‬‭an‬‭integral‬‭part‬‭of‬‭m‬‭and‬‭if‬‭m‬‭is‬‭an‬‭integer,‬‭then‬‭there‬‭are‬‭two‬
‭modes one at (m-1) and the other at m and the probabilities in both these cases are equal. [1]‬

‭3.3 Parameters of Poisson Distribution‬

‭3.3.1 Moments About Origin -‬


∞ ‭𝑥‬
‭'‬ −‭𝑚‬ ‭𝑚‬
‭[Link] First moment about the origin -‬µ‭1‬ = ∑ ‭𝑒‬ ‭𝑥!‬
‭𝑥‬ = ‭𝑚‬
‭𝑥=
‬ ‭0‬

∞ ‭𝑥‬
‭'‬ −‭𝑚‬ ‭𝑚‬ ‭2‬ ‭2‬
‭[Link] Second moment about the origin -‬µ‭2‬ = ∑ ‭𝑒‬ ‭𝑥!‬
‭𝑥‬ = ‭𝑚‬ + ‭𝑚‬
‭𝑥=
‬ ‭0‬

∞ ‭𝑥‬
‭'‬ −‭𝑚‬ ‭𝑚‬ ‭2‬ ‭3‬ ‭2‬
‭[Link] Third moment about the origin -‬µ‭3‬ = ∑ ‭𝑒‬ ‭𝑥!‬
‭𝑥‬ = ‭𝑚‬ + ‭3‭𝑚
‬ ‬ + ‭𝑚‬
‭𝑥=
‬ ‭0‬

∞ ‭𝑥‬
‭'‬ −‭𝑚‬ ‭𝑚‬ ‭2‬ ‭4‬ ‭3‬ ‭2‬
‭[Link] Fourth moment about the origin -‬µ‭4‬ = ∑ ‭𝑒‬ ‭𝑥!‬
‭𝑥‬ = ‭𝑚‬ + ‭6‭𝑚
‬ ‬ + ‭7‭𝑚
‬ ‬ + ‭𝑚‬
‭𝑥=
‬ ‭0‬

‭3.3.2 Moments about Mean‬

‭[Link] First moment about mean -‬µ‭1‬‭‬ = ‭0‬

‭'‬ ‭'2‬
‭[Link] Second moment about mean -‬µ‭2‬‭‬ = µ‭2‬ − ‭µ
‬ = ‭𝑚‬

‭Variance‬‭=‬µ‭2‬‭‬ = ‭𝑚‬‭and‬‭Standard Deviation‬‭=‬ ‭𝑚‬

‭'‬ ‭'‬ ‭'‬ ‭'3‬


‭[Link] Third moment about mean -‬µ‭3‬ = µ‭3‬ − ‭3‬‭µ
‬ ‭2‬µ‭1‬ + ‭2µ
‬ ‭1‬ = ‭𝑚‬

‭'‬ ‭'‬ ‭'‬ ‭'‬ ‭'2‬ ‭'4‬ ‭2‬


‭[Link] Fourth moment about the mean‬‭-‬µ‭4‬ = µ‭4‬ − ‭4‬‭µ
‬ ‭3‬µ‭1‬ + ‭6µ
‬ ‭2‬µ‭1‬ − ‭3µ
‬ ‭1‬ = ‭3‭𝑚
‬ ‬ + ‭𝑚‬

‭Note:‬‭The graphical representation of Poisson Distribution. [2]‬

‭Figure 1.3.‬
‭𝑡‬
‭𝑡𝑋‬ (‭𝑚‬(‭𝑒‬ −‭1‬))
‭3.3‬ ‭Moment‬ ‭Generating‬ ‭Function‬ ‭-‬ ‭𝑀‬‭𝑋‭𝑡‬ ‬‭‬ = ‭‬‭𝐸(‬ ‭𝑒‬ ) = ‭𝑒‬ ‭,‬ ‭where‬ ‭m‬ ‭is‬ ‭the‬ ‭mean‬ ‭(and‬
‭variance) of the Poisson Distribution.‬

‭3.4 Pearson’s coefficient -‬


‭2‬
µ‭3‬ ‭2‬
‭𝑚‬
β‭1‬ = ‭3‬ = ‭3‬
µ‭2‬ ‭𝑚‬

‭1‬
‭or,‬ β‭1‬ = ‭ ‬
𝑚

‭1‬
‭Kurtosis =‬β‭1‬ = ‭ ‬
𝑚

µ‭4‬ ‭2‬
‭3‬‭𝑚‬ +‭𝑚‬
β‭2‬ = ‭2‬ = ‭2‬
µ‭2‬ ‭𝑚‬

‭1‬
‭or,‬ β‭1‬ = ‭3‬ + ‭ ‬
𝑚

‭1‬
γ‭1‬ = ‭‬ β‭1‬ =
‭𝑚‬

‭1‬
‭Skewness =‬γ‭1‬ =
‭𝑚‬

‭1‬
γ‭2‬ = β‭2‬ − ‭3‬ = ‭ ‬
𝑚

‭ ote:‬‭A‬‭poison‬‭distribution‬‭is‬‭positively‬‭skewed.‬‭However,‬‭the‬‭degree‬‭of‬‭skewness‬‭decreases‬‭with‬‭an‬
N
‭increase in the value of λ. See in figure 1.3.. Here λ = m (parameter). [2]‬

‭Figure 1.4.‬

‭4 Methodologies Utilizing Poisson Distribution‬

‭ he‬ ‭Poisson‬ ‭distribution‬ ‭is‬ ‭commonly‬ ‭used‬ ‭to‬ ‭model‬‭the‬‭probability‬‭of‬‭a‬‭specific‬‭number‬‭of‬‭events‬


T
‭occurring‬ ‭within‬ ‭a‬ ‭fixed‬ ‭interval‬ ‭of‬ ‭time‬ ‭or‬ ‭space.‬ ‭This‬ ‭is‬ ‭applicable‬ ‭when‬ ‭the‬ ‭events‬ ‭happen‬
‭independently‬ ‭of‬ ‭one‬ ‭another‬ ‭and‬ ‭at‬ ‭a‬ ‭constant‬ ‭average‬ ‭rate.‬ ‭Below‬ ‭are‬ ‭some‬ ‭methodologies‬ ‭and‬
‭approaches that are frequently associated with the Poisson distribution:‬
‭(These methodologies are referred from [2], [4] and [6]).‬
‭4.1 Parameter Estimation‬
‭●‬ ‭Maximum‬ ‭Likelihood‬ ‭Estimation‬ ‭(MLE):‬ ‭For‬ ‭a‬ ‭Poisson‬ ‭distribution‬ ‭with‬ ‭parameter‬ ‭m‬
‭𝑛‬
‭‬
1
‭(average rate of events), MLE estimates m as the sample mean,‬‭𝑚‬= ‭𝑛‬
∑ ‭𝑥‭𝑖‬‬.
‭𝑖‬=‭1‬
‭●‬ B‭ ayesian‬‭Estimation:‬‭Assumes‬‭a‬‭prior‬‭distribution‬‭for‬‭m‬‭(often‬‭Gamma‬‭distribution‬‭as‬‭it‬‭is‬
‭conjugate‬‭to‬‭Poisson),‬‭then‬‭combines‬‭it‬‭with‬‭observed‬‭data‬‭to‬‭form‬‭a‬‭posterior‬‭distribution‬‭for‬
‭m.‬
‭4.2 Goodness of Fit Testing‬

‭●‬ C ‭ hi-square‬ ‭Test:‬ ‭Used‬ ‭to‬ ‭determine‬ ‭if‬ ‭observed‬ ‭frequencies‬ ‭match‬ ‭expected‬ ‭frequencies‬
‭under the Poisson model.‬
‭●‬ ‭Likelihood‬ ‭Ratio‬ ‭Test‬ ‭(LRT):‬ ‭Compares‬ ‭the‬ ‭likelihood‬ ‭of‬ ‭data‬ ‭under‬ ‭a‬ ‭Poisson‬ ‭model‬
‭against an alternative hypothesis model.‬

‭4.3 Modelling Overdispersion‬

‭●‬ W ‭ hen‬ ‭data‬ ‭show‬ ‭more‬ ‭variability‬ ‭than‬ ‭a‬ ‭Poisson‬ ‭model‬ ‭predicts,‬ ‭Negative‬ ‭Binomial‬
‭Distribution‬‭or‬‭Quasi-Poisson models‬‭are often used.‬
‭●‬ ‭Zero-inflated Poisson (ZIP)‬‭and‬‭Hurdle Models‬‭help model data with excess zeros.‬

‭4.4 Regression Analysis with Poisson Models‬

‭●‬ P ‭ oisson‬‭Regression:‬‭For‬‭count‬‭data‬‭influenced‬‭by‬‭predictor‬‭variables,‬‭the‬‭log‬‭of‬‭the‬‭expected‬
‭count is modelled as a linear combination of the predictors.‬
‭●‬ ‭Generalized‬ ‭Linear‬ ‭Models‬ ‭(GLMs):‬ ‭Poisson‬ ‭GLMs‬ ‭use‬ ‭a‬ ‭log-link‬ ‭function,‬ ‭enabling‬
‭regression analysis for count data.‬

‭4.5 Time Series and Spatial Poisson Models‬

‭‬ P
● ‭ oisson Processes:‬‭Continuous-time processes used‬‭to model counts over time or space.‬
‭●‬ ‭Spatial‬ ‭Poisson‬ ‭Models:‬ ‭Used‬ ‭for‬ ‭modelling‬ ‭spatial‬ ‭data‬ ‭where‬ ‭the‬ ‭location‬ ‭of‬ ‭events‬ ‭is‬
‭random and follows a Poisson process.‬

‭4.6 Simulation Techniques‬

‭●‬ M ‭ onte‬ ‭Carlo‬ ‭Simulation:‬ ‭Generate‬ ‭synthetic‬ ‭data‬ ‭to‬ ‭understand‬ ‭and‬ ‭validate‬ ‭Poisson‬
‭distribution models.‬
‭●‬ ‭Bootstrapping:‬ ‭For‬ ‭generating‬ ‭confidence‬ ‭intervals‬ ‭or‬ ‭distributions‬ ‭of‬ ‭Poisson-based‬
‭estimates.‬

‭ hese‬‭methodologies‬‭offer‬‭flexibility‬‭in‬‭modelling‬‭different‬‭scenarios‬‭using‬‭the‬‭Poisson‬‭distribution.‬
T
‭They‬ ‭ensure‬ ‭that‬ ‭the‬ ‭approach‬ ‭can‬ ‭adapt‬ ‭to‬ ‭data‬ ‭exhibiting‬ ‭extra‬ ‭variability,‬ ‭dependencies‬ ‭on‬
‭predictors,‬ ‭or‬ ‭specific‬ ‭spatial-temporal‬ ‭patterns.‬ ‭The‬ ‭choice‬ ‭of‬ ‭each‬ ‭technique‬ ‭depends‬ ‭on‬ ‭the‬ ‭data‬
‭structure, underlying assumptions, and the specific requirements of the research or application.‬

‭5 POISSON DISTRIBUTION AS A LIMITING FORM OF BINOMIAL DISTRIBUTION‬


‭Statement:‬ ‭Poisson‬ ‭Distribution‬ ‭is‬ ‭the‬ ‭limiting‬ ‭form‬ ‭of‬ ‭the‬ ‭binomial‬ ‭distribution‬
‭𝑛‬
(‭𝑞‬ + ‭𝑝‬) ‭‬‭𝑤ℎ𝑒𝑛‬‭‬‭𝑝‭‬‬(‭‭𝑜 ‬ 𝑛𝑑‬‭‬‭𝑛‬→ ∞, ‭such‬‭that‬‭np‬‭(‬‭or‬‭nq)‬‭is‬‭a‬‭finite‬‭quantity‬‭say‬‭m‬‭i.e.,‬‭np‬‭=‬
‬ 𝑟‬‭‬‭𝑞)‬ ‭‬→ ‭0‬‭‭𝑎
‭m. (See Figure 1.5.) [1]‬

‭Figure 1.5.‬

‭6‬‭Benchmark Problems Employing Poisson Distribution‬

‭(These examples have been referred from [1])‬

‭6.1.‬‭What is the probability of throwing an ace with a die exactly 10 times in 1200 trials?‬

‭Solution. The probability p of throwing an ace with a die = (⅙)‬

‭Also number of trials = n = 1200‬

‭Therefore, if the Poisson distribution is used then the mean m = np = 1200 * (⅙) = 200.‬
−‭𝑚‬ ‭10‬
‭𝑒‬ ‭𝑚‬
‭Therefore, the required probability of throwing an ace is exactly 10 times =‬ (‭10‬)!
, ‭‬‭where m = 200.‬

−‭200‬ ‭10‬
‭Therefore, probability is = [‬‭𝑒‬ * (‭200‬) ]‭/‬(‭10‬)!

‭ .2.‬‭A‬‭manufacturer‬‭of‬‭screws‬‭knows‬‭that‬‭5%‬‭of‬‭his‬‭product‬‭is‬‭defective.‬‭If‬‭he‬‭sells‬‭them‬‭in‬‭boxes‬‭of‬
6
‭100‬‭and‬‭guarantees‬‭that‬‭no‬‭more‬‭than‬‭10‬‭screws‬‭will‬‭be‬‭defective,‬‭what‬‭is‬‭the‬‭approximate‬‭probability‬
‭that a box will fail to meet the guaranteed quality?‬

‭Solution. Here p = (5/100); n = 100 so m = np = 100* (5/100) = 5.‬

‭ lso,‬ ‭we‬ ‭know‬ ‭that‬ ‭according‬ ‭to‬ ‭Poisson‬ ‭distribution‬ ‭(here‬ ‭p‬ ‭is‬ ‭very‬ ‭small),‬ ‭the‬ ‭probability‬ ‭of‬ ‭x‬
A
‭screws being defective is given by‬
−‭𝑚‬ ‭𝑥‬ −‭5‬ ‭𝑥‬
‭𝑒‬ ‭𝑚‬ ‭𝑒‬ ‭5‬
(‭𝑥)‬ !
‭i.e.,‬ (‭𝑥)‬ !
‭10‬ −‭5‬ ‭𝑥‬
‭𝑒‬ ‭5‬
‭Therefore, the probability of ten or less than ten screws to be defective =‬ ∑ (‭𝑥)‬ !
‭.‬
‭𝑥‬=‭0‬

‭10‬ −‭5‬ ‭𝑥‬


‭𝑒‬ ‭5‬
‭Therefore, the required probability of the defective screws to be more than ten = 1 -‬ ∑ (‭𝑥)‬ !
‭.‬
‭𝑥‬=‭0‬

‭7 APPLICATIONS‬

‭ he‬‭Poisson‬‭distribution‬‭is‬‭a‬‭statistical‬‭model‬‭for‬‭describing‬‭random‬‭and‬‭independent‬‭events‬‭within‬‭a‬
T
‭fixed timeframe or area. Here are key applications across various fields:‬

‭These applications are referred from [4], [5] and [6].‬

‭7.1 Telecommunications‬

‭ .‬ ‭Call‬ ‭Centers:‬ ‭The‬ ‭number‬ ‭of‬ ‭calls‬ ‭received‬ ‭at‬ ‭a‬ ‭call‬ ‭centre‬ ‭per‬ ‭hour‬ ‭often‬ ‭follows‬ ‭a‬ ‭Poisson‬
1
‭distribution, which helps optimize staffing.‬

‭ .‬‭Network‬‭Traffic:‬‭The‬‭arrival‬‭of‬‭data‬‭packets‬‭at‬‭routers‬‭can‬‭be‬‭modelled‬‭as‬‭a‬‭Poisson‬‭process,‬‭aiding‬
2
‭in congestion management.‬

‭7.2 Healthcare and Epidemiology‬

‭ .‬ ‭Disease‬ ‭Incidence:‬ ‭New‬ ‭cases‬ ‭of‬ ‭a‬ ‭rare‬ ‭disease‬ ‭within‬ ‭a‬ ‭specific‬ ‭timeframe‬ ‭typically‬ ‭follow‬ ‭a‬
1
‭Poisson distribution, assisting in predicting outbreaks.‬

‭ .‬ ‭Emergency‬ ‭Room‬ ‭Demand:‬ ‭Hospitals‬ ‭use‬ ‭the‬ ‭model‬‭to‬‭estimate‬‭daily‬‭patient‬‭arrivals,‬‭helping‬‭in‬


2
‭resource allocation.‬

‭7.3 Environmental Science‬

‭ .‬‭Natural‬‭Events:‬‭Rare‬‭events‬‭like‬‭earthquakes‬‭can‬‭be‬‭modelled‬‭using‬‭Poisson‬‭distribution‬‭to‬‭predict‬
1
‭occurrence likelihoods.‬

‭2. Wildlife Conservation: Tracking sightings of rare species helps in conservation assessments.‬

‭7.4 Manufacturing and Quality Control‬

‭ .‬ ‭Defect‬ ‭Counts:‬ ‭The‬ ‭number‬ ‭of‬ ‭defects‬ ‭in‬‭a‬‭batch‬‭can‬‭be‬‭estimated‬‭using‬‭Poisson‬‭distribution‬‭for‬


1
‭quality control.‬

‭2. Machine Failures: Equipment breakdowns over time can also follow this distribution.‬

‭7.5 Finance and Insurance‬

‭1. Claim Counts: Insurance companies model claim filings using Poisson distributions to assess risks.‬

‭ .‬ ‭Bank‬ ‭Transactions:‬ ‭The‬ ‭number‬ ‭of‬ ‭daily‬ ‭transactions‬ ‭can‬ ‭follow‬ ‭a‬ ‭Poisson‬ ‭process‬ ‭for‬ ‭cash‬
2
‭management.‬
‭7.6 Transportation and Traffic Flow‬

‭ .‬ ‭Accidents:‬ ‭The‬ ‭occurrence‬ ‭of‬ ‭traffic‬ ‭incidents‬ ‭often‬ ‭aligns‬ ‭with‬ ‭a‬ ‭Poisson‬ ‭distribution,‬ ‭helping‬
1
‭improve safety measures.‬

‭ .‬‭Bus‬‭Arrivals:‬‭The‬‭number‬‭of‬‭arrivals‬‭at‬‭bus‬‭stops‬‭modelled‬‭as‬‭a‬‭Poisson‬‭process‬‭aids‬‭in‬‭optimizing‬
2
‭schedules.‬

‭7.7 Crime and Public Safety‬

‭ .‬ ‭Crime‬ ‭Incidents:‬ ‭Police‬ ‭use‬ ‭Poisson‬ ‭distribution‬ ‭to‬ ‭model‬ ‭crime‬ ‭occurrences,‬ ‭aiding‬ ‭in‬ ‭resource‬
1
‭allocation.‬

‭ .‬‭Fire‬‭Calls:‬‭The‬‭number‬‭of‬‭fire‬‭incidents‬‭is‬‭frequently‬‭modelled‬‭using‬‭this‬‭distribution‬‭for‬‭effective‬
2
‭resource management.‬

‭7.8 Sports and Gaming‬

‭ .‬‭Goals‬‭in‬‭Soccer‬‭Matches:‬‭The‬‭number‬‭of‬‭goals‬‭scored‬‭can‬‭follow‬‭a‬‭Poisson‬‭distribution,‬‭assisting‬
1
‭in predicting outcomes.‬

‭ .‬ ‭Player‬ ‭Performance:‬ ‭Specific‬ ‭events‬ ‭in‬ ‭games,‬ ‭like‬ ‭three-pointers,‬ ‭can‬ ‭also‬ ‭be‬ ‭modelled‬ ‭using‬
2
‭Poisson.‬

‭7.9 Biology and Genetics‬

‭1. Mutations: Genetic mutations during cell division can be modelled using the Poisson distribution.‬

‭2. Cell Biology: Tracking cell division or death in experiments often uses this statistical approach.‬

‭8 CONCLUSION AND FUTURE PROSPECTS‬

I‭ n‬‭conclusion,‬‭the‬‭Poisson‬‭distribution‬‭is‬‭a‬‭valuable‬‭tool‬‭for‬‭predicting‬‭rare‬‭or‬‭random‬‭events‬‭in‬‭many‬
‭fields,‬ ‭such‬ ‭as‬ ‭healthcare,‬ ‭environmental‬ ‭science,‬ ‭manufacturing,‬ ‭and‬ ‭public‬ ‭safety.‬ ‭It‬ ‭works‬ ‭well‬
‭because‬‭it‬‭captures‬‭how‬‭often‬‭events‬‭happen‬‭over‬‭set‬‭periods.‬‭This‬‭makes‬‭it‬‭ideal‬‭for‬‭events‬‭that‬‭occur‬
‭independently and at a steady average rate.‬

‭ y‬ ‭using‬ ‭Poisson‬ ‭models,‬ ‭organizations‬ ‭and‬ ‭researchers‬ ‭can‬ ‭allocate‬ ‭resources‬ ‭better,‬ ‭streamline‬
B
‭processes,‬ ‭and‬ ‭make‬ ‭smart‬ ‭decisions.‬ ‭This‬ ‭helps‬ ‭them‬ ‭meet‬ ‭demand,‬ ‭improve‬ ‭safety,‬ ‭and‬ ‭boost‬
‭efficiency.‬ ‭The‬ ‭Poisson‬ ‭distribution‬ ‭is‬ ‭essential‬ ‭for‬‭analyzing‬‭the‬‭occurrence‬‭of‬‭random‬‭events‬‭over‬
‭time‬ ‭or‬ ‭space.‬ ‭It‬ ‭provides‬ ‭a‬ ‭clear‬ ‭way‬ ‭to‬ ‭assess‬ ‭how‬ ‭likely‬ ‭certain‬ ‭counts‬ ‭are‬ ‭when‬ ‭these‬ ‭events‬
‭happen independently and at a regular rate.‬

‭ he‬ ‭future‬ ‭of‬ ‭the‬ ‭Poisson‬ ‭distribution‬ ‭appears‬ ‭promising,‬ ‭thanks‬ ‭to‬ ‭its‬ ‭growing‬ ‭integration‬ ‭with‬
T
‭advanced‬ ‭analytics,‬ ‭machine‬ ‭learning,‬ ‭and‬ ‭real-time‬ ‭data‬ ‭applications.‬ ‭As‬ ‭technology‬ ‭and‬ ‭data‬
‭generation‬ ‭continue‬ ‭to‬ ‭accelerate,‬ ‭the‬ ‭Poisson‬ ‭model‬ ‭is‬ ‭expected‬ ‭to‬ ‭evolve‬ ‭significantly,‬ ‭enabling‬
‭complex decision-making, predictive analysis, and automation. .‬

I‭ ts‬ ‭simplicity‬ ‭and‬ ‭flexibility‬ ‭make‬ ‭it‬ ‭especially‬ ‭helpful‬ ‭for‬ ‭counting‬ ‭data.‬ ‭This‬ ‭includes‬ ‭tracking‬
‭disease‬‭cases,‬‭managing‬‭service‬‭resources,‬‭or‬‭studying‬‭rare‬‭events‬‭in‬‭research.‬‭Ultimately,‬‭the‬‭Poisson‬
‭ istribution‬‭connects‬‭theoretical‬‭probability‬‭with‬‭real-life‬‭applications.‬‭It‬‭gives‬‭a‬‭solid‬‭foundation‬‭for‬
d
‭data-driven planning and forecasting.‬

‭REFERENCES‬

[‭1]‬ ‭S.‬ ‭Gupta‬ ‭and‬ ‭V.‬ ‭Kapoor,‬ ‭Fundamentals‬ ‭of‬ ‭Mathematical‬ ‭Statistics‬‭.‬ ‭Sultan‬ ‭Chand‬ ‭&‬ ‭Sons,‬
‭2020.‬
‭[2]‬ ‭“HELM‬ ‭Workbook‬ ‭37‬ ‭Discrete‬ ‭Probability‬ ‭Distributions.”‬ ‭[Link] ‭2008.‬
‭[Link] (accessed: 2008).‬
‭[3]‬ ‭“UNIT‬ ‭10‬ ‭POISSON‬ ‭DISTRIBUTION.”‬ ‭eGyanKosh.‬
‭[Link] (accessed: Nov. 05, 2024).‬
‭[4]‬ ‭“An‬ ‭Extension‬ ‭of‬ ‭the‬ ‭Poisson‬ ‭Distribution:‬ ‭Features‬ ‭and‬ ‭Application‬ ‭for‬ ‭Medical‬ ‭Data‬
‭Modeling.” MDPI. Apr. 28, 2023.‬‭[Link]
‭[5]‬ ‭“A‬ ‭Short‬ ‭Review‬ ‭on‬ ‭Poisson‬ ‭Distribution‬ ‭with‬ ‭Some‬ ‭Applications‬ ‭and‬ ‭..”‬ ‭[Link].‬
‭[Link] (accessed: Nov. 06, 2024).‬
‭[6]‬ ‭“The‬ ‭Properties‬ ‭and‬ ‭Application‬ ‭of‬ ‭Poisson‬ ‭Distribution‬ ‭-‬ ‭IOPscience.”‬ ‭[Link].‬
‭[Link] (accessed: Nov. 06, 2024).‬

You might also like